ai-ugc-factory
Pass
Audited by Gen Agent Trust Hub on Nov 16, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill references external resources for installation and setup. It points to the author's own repository for installation instructions and suggests using tools from a well-known and trusted organization (Vercel Labs).
- [INDIRECT_PROMPT_INJECTION]: The skill identifies an attack surface where the agent processes external content from browsers and media metadata. However, it provides robust instructions to mitigate this risk.
- Ingestion points: Web page content accessed via browser tools (referenced in
references/browser-runtime-and-sessions.md) and media transcripts/metadata processed by media tools. - Boundary markers: The instructions explicitly warn the agent to treat page text, tool descriptions, and error messages as data only, rather than instructions or permissions to disclose secrets or run commands.
- Capability inventory: The skill uses browser tools for interaction and
ffmpeg/ffprobefor media manipulation. - Sanitization: The skill mandates verifying actual form fields and destinations before data submission, ensuring the agent does not blindly follow instructions embedded in web pages.
Audit Metadata